Seite - 195 - in Proceedings - OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“

TwistMessages Odometry Messages Interface OdometryMessages AckermannMessages SerialAckerman Messages Serial Odometry Messages RealCar SimulatedCar Figure1: The interactionof thedifferentmessagesof the interface. 4. MotionModel Simple problems like wheel slips, bumps, and inaccuracies within the robot effect the robots mo- tion [4, 7]. The tracking system for this project is based on the velocity motion model which consid- ers theseerrors. Although thevelocitymotionmodel introducedbyThrun[4] is conceived for robots able to turn around their own axis, its simple structure allows its customization for ackermann drive robots. Based on the motion commands, the velocity motion model calculates the robots estimated pose and a matrix in which its uncertainty is contained. This is equal to the prediction step of a Kalmanfilter [4, 7]. Further, themotionmodel isapplied toaflat space representedbyxandy and theparameterθwhich stands for the robotsorientationas shown inFigure2. w he el ba se ϕ R pWr v θW yW xWpW0 track kingpin ϕR ϕL Figure2: Thegeometryof theackermann robot in the twodimensional space. 195